Prediction of body weights of small mammals from skull measurements

Abstract

Prediction oí body weights of small mammals from skull measurements. Acta theriol., 29, 31: 367—381 [With 6 Tables & 2 Figs.]. Body weights of small mammals may be predicted with accuracy from measurements of cranial or mandibular dimensions. Stepwise multiple regression techniques were used to generate "best" equations (those with largest coefficients of determination). "Best" predictive equations obtained from cranial measurements usually contained three independent variables, the identity of which varied among species. Equations using only mandibular measurements were less accurate than those based on cranial measurements. In order to obtain predictive equations with maximum accuracy, sample sizes should be at least N=40. Logarithmic transformations may improve the predictive ac-curacy of the equations, but only slightly. Equations such as these may be useful whenever skull measurements can be made but body weights are unavailable. Specifically, weights may be estimated from skeletal elements in pellets regurgitated by owls. From these remains one can use multiple regression equations to determine: (1) the impact of the predator on specific size classes within single mammalian prey species, (2) relative frequency of occurrence of mammalian prey of various sizes, and (3) energy budgets of owls and rates of secondary production.